FeedAIde: Guiding App Users to Submit Rich Feedback Reports by Asking Context-Aware Follow-Up Questions
New AI framework captures app screenshots to ask context-aware follow-ups, improving bug report completeness.
A team of researchers has introduced FeedAIde, a novel framework designed to bridge the gap between vague user feedback and the detailed information developers need. Presented in a paper accepted at MOBILESoft 2026, the system tackles the chronic problem of incomplete bug reports and feature requests by leveraging the reasoning power of Multimodal Large Language Models (MLLMs). Instead of a static form, FeedAIde interactively guides users, starting by capturing a screenshot of the app issue to establish crucial context. This visual data allows the AI to ask intelligent, adaptive follow-up questions, collaboratively refining a user's initial report into a comprehensive document.
The researchers implemented FeedAIde as an iOS framework and conducted a real-world evaluation using a gym application. Compared to the app's standard feedback form, participants found the AI-powered tool significantly easier and more helpful for reporting issues. An expert assessment of the 54 generated reports confirmed a marked improvement in quality, particularly in completeness—a key metric for developer efficiency. This study demonstrates the practical potential of context-aware, GenAI-powered interfaces to simultaneously enhance user experience and deliver high-value, actionable feedback, potentially saving development teams countless hours previously spent on clarification cycles.
- Uses Multimodal LLMs (MLLMs) to analyze app screenshots for context-aware questioning.
- Evaluated on a real iOS gym app, generating 54 reports rated more complete by experts.
- Accepted for publication at the MOBILESoft 2026 conference, indicating peer-reviewed academic rigor.
Why It Matters
Transforms vague user complaints into structured, developer-ready reports, drastically reducing back-and-forth clarification time.